if apple's flailing around, not sure what to do, it seems to me that a decent placeholder would be: - for each paying icloud customer - run a small LLM model

if apple's flailing around, not sure what to do, it seems to me that a decent placeholder would be: - for each paying icloud customer - run a small LLM model (llama or gemma3) that is fine-tuned on that customer's cloud content - provide access to the model in a flashy new app which lets them ask questions about, search and generate derivative works from their own content that's it i bet a lot of people would like it